What are the limits of efficiency for the application of bi-cubic splines in a regional weather prediction grid point model?

Single (one-dimensional) and bi-cubic (two-dimensional) splines have been used in numerical weather prediction models since about 1970. Especially the semi-Lagrangian advection schemes have been developed with the help of the spline interpolation method. In this paper the consequent formulation of a...

Full description

Bibliographic Details
Main Authors: Thomas Prenosil, H.G. Becker
Format: Article
Language:English
Published: Borntraeger 2000-01-01
Series:Meteorologische Zeitschrift
Subjects:
Online Access:http://dx.doi.org/10.1127/metz/8/2000/168
Description
Summary:Single (one-dimensional) and bi-cubic (two-dimensional) splines have been used in numerical weather prediction models since about 1970. Especially the semi-Lagrangian advection schemes have been developed with the help of the spline interpolation method. In this paper the consequent formulation of a primitive equation grid point model with bi-cubic splines is investigated. Numerical experiments show that the complete replacement of all horizontal finite difference approximations in the advective and non-advective terms by the associated bi-cubic splines leads to a minor increase of forecast quality, which is accompanied by an inappropriate and uneconomical increase of computing time. With respect to the semi-Lagrangian treatment of the horizontal advection problem, split explicit schemes with single cubic splines produce a slightly reduced forecast quality compared with the more complex bi-cubic splines. Because on the other hand the numerical costs are not raised by the two-dimensional advection method, the use of the physically more reasonable bi-cubic spline interpolation is recommended.
ISSN:0941-2948